COAR has published a paper presenting a multilingual semantic search approach for scholarly discovery and has opened a feedback period from November 6 to December 19, 2025. The paper has outlined an approach that retrieves information by meaning rather than by exact wording, enabling users to search in any supported language and discover relevant work across other languages. The method is not a translation technology but a way to associate meaning with related content across different languages.
Instead of specifying a fixed technical design, the document has invited the community to consider how the model could evolve within the broader ecosystem of scholarly communication, reflecting on early experiences, shared principles, and collective responsibilities aimed at advancing openness, equity, and linguistic diversity.
To accompany the paper, COAR has released a demonstration notebook that openly shares the code for a simple proof of concept. The notebook has illustrated how multilingual embeddings can represent meaning across languages and enable cross-language discovery. It has been presented as a minimal, open demonstration of the underlying technology intended to encourage learning, experimentation, and community-based development rather than as a full application.
COAR has noted that the global repository ecosystem is extremely diverse and multilingual and has indicated that it is investigating how this approach could significantly improve access to content in other languages within the repository network. COAR has requested input from members and the broader community on the usefulness and feasibility of the semantic multilingual search model as applied to the global repository network and the scholarly commons in general. Based on the feedback, COAR will consider the possibility of launching a project to investigate how best to apply this approach in the repository network.
The paper will be available in Spanish and Portuguese shortly.
Click here to read the original press release.
